Introduction

Investigating how health behaviours change during pregnancy.

  • How these differ between maternal and paternal behaviours

    • Reflects differences in health advice given to each

  • Correlations between mum and dads behaviour

    • Across different health behaviours
    • Shows how interelated different behaviours are
    • Are correlations higher if live with partner (assortative mating and shared environment)

  • Patterns stratified by if partner also stops behaviour

  • Association between SEP and health behaviours

    • See how socially patterened these behaviours are

  • Do behaviours return to baseline postnatally

    • If slightly lower, this is a good time to intervene to improve health
    • Also shows where support may be needed to help mothers maintain healthier behaviours after birth

  • How this differs between cohorts

    • Highlighting different confounding structures in different populations
    • Implications for multi-cohort studies of maternal/paternal effects on child health

Using variables derived previously within EPoCH

Prepare dataset

Load packages

library("tidyverse")
## ── Attaching packages ─────────────────────────────────────── tidyverse 1.3.0 ──
## ✓ ggplot2 3.3.3     ✓ purrr   0.3.4
## ✓ tibble  3.1.1     ✓ dplyr   1.0.6
## ✓ tidyr   1.1.2     ✓ stringr 1.4.0
## ✓ readr   1.4.0     ✓ forcats 0.5.0
## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## x dplyr::filter() masks stats::filter()
## x dplyr::lag()    masks stats::lag()

ALSPAC

Load data

This is taken from the ALSPAC dataset created for EPoCH

dat <- readRDS("/Volumes/MRC-IEU-research/projects/ieu2/p5/015/working/data/alspac/alspac_pheno.rds")

Select variables

vars<-
  c(
    
    "physact_mother_secondtrim_ordinal",
    "physact_mother_thirdtrim_ordinal",
    "physact_father_secondtrim_ordinal",
     # "physact_mother_postnatal_ordinal", (not included as not directly comparable)
    
    "caffeine_mother_total_firsttrim_ordinal",
   "caffeine_mother_total_secondtrim_ordinal",
    "caffeine_mother_total_thirdtrim_ordinal",
    "caffeine_mother_total_postnatal_ordinal",
    "caffeine_father_total_secondtrim_ordinal",
    
    "smoking_mother_preconception_ordinal",
    "smoking_mother_firsttrim_ordinal",
    "smoking_mother_secondtrim_ordinal",
    "smoking_mother_thirdtrim_ordinal",
    "smoking_mother_postnatal_ordinal",
    
    "smoking_father_startpregnancy_ordinal",
    "smoking_father_secondtrim_ordinal",
    "smoking_father_thirdtrim_ordinal",
    "smoking_father_postnatal_ordinal",
    
    "alcohol_mother_preconception_ordinal",
    "alcohol_mother_firsttrim_ordinal",
    "alcohol_mother_secondtrim_ordinal",
    "alcohol_mother_thirdtrim_ordinal",
    "alcohol_mother_postnatal_ordinal",
    
    "alcohol_father_preconception_ordinal",
    "alcohol_father_secondtrim_ordinal",
    "alcohol_father_thirdtrim_ordinal",
    "alcohol_father_postnatal_ordinal",
    
    "covs_occup_mother",
    "covs_occup_father",
    
    "covs_edu_mother",
    "covs_edu_father",
    
    "covs_occup_mother_highestlowest_binary",
    "covs_occup_father_highestlowest_binary",
    "covs_edu_mother_highestlowest_binary",
    "covs_edu_father_highestlowest_binary",
    
    "covs_parity_mother_binary",
    "covs_partner_lives_with_mother_prenatal"

  )

Correlations

! To do format plots

Parents correlation

This shows how maternal and paternal health behaviours are correlated with each other and SEP

#install.packages("corrplot")

require(dplyr)
require(corrplot)
## Loading required package: corrplot
## corrplot 0.89 loaded
df1<-mutate_all(dat[,vars], function(x) as.numeric(x))
cors <- cor(df1,use="pairwise.complete")
corrplot(cors,tl.cex=0.5,method="color",tl.col="black",order="hclust",addrect=5)

Parents change over time

This shows how parental behaviour changes during and after pregnancy

require(reshape2)
## Loading required package: reshape2
## 
## Attaching package: 'reshape2'
## The following object is masked from 'package:tidyr':
## 
##     smiths
df1<-mutate_all(dat[,c("aln",vars)], function(x) as.numeric(x))
df1_melt <- melt(df1[,grepl("aln|smoking|physact|caff|alco",names(df1))],id.vars = "aln")

df1_melt$time <- NA
df1_melt$time[grepl(df1_melt$variable,pattern="preconception")]<-"pre-conception"
df1_melt$time[grepl(df1_melt$variable,pattern="firsttrim|start")]<-"first trimester"
df1_melt$time[grepl(df1_melt$variable,pattern="secondtrim")]<-"second trimester"
df1_melt$time[grepl(df1_melt$variable,pattern="thirdtrim")]<-"third trimester"
df1_melt$time[grepl(df1_melt$variable,pattern="postnatal")]<-"postnatal"
df1_melt$time <- factor(df1_melt$time,ordered=T, levels=c("pre-conception","first trimester","second trimester","third trimester","postnatal"))

df1_melt$behaviour <- NA
df1_melt$behaviour[grepl(df1_melt$variable,pattern="caff")]<-"caffeine"
df1_melt$behaviour[grepl(df1_melt$variable,pattern="alcohol")]<-"alcohol"
df1_melt$behaviour[grepl(df1_melt$variable,pattern="smoking")]<-"smoking"
df1_melt$behaviour[grepl(df1_melt$variable,pattern="physac")]<-"physical activity"
df1_melt$behaviour <- factor(df1_melt$behaviour,ordered=T, levels=c("smoking","alcohol","caffeine","physical activity"))

df1_melt$parent <- NA
df1_melt$parent[grepl(df1_melt$variable,pattern="mother")]<-"mother"
df1_melt$parent[grepl(df1_melt$variable,pattern="father")]<-"partner"

x <- droplevels(df1_melt[order(df1_melt$aln,df1_melt$time,df1_melt$parent,df1_melt$behaviour),])
x<-x[is.na(x$value)==F,]
x <- x[duplicated(x[,c("parent","behaviour","aln")])==F,]

names(x)[3]<-"initial_dose"
names(x)[2]<-"initial_var"

dfx <- merge(x[,c("aln","parent","behaviour","initial_dose","initial_var")],df1_melt,by=c("aln","parent","behaviour"),all=T)

dfy <- 
  dfx %>%
  group_by(factor(initial_dose),factor(behaviour),factor(time),factor(parent)) %>%
  dplyr::summarise(mean=mean(value,na.rm=T),sd=sd(value,na.rm=T),n=n())
## `summarise()` has grouped output by 'factor(initial_dose)', 'factor(behaviour)', 'factor(time)'. You can override using the `.groups` argument.
ggthemr::ggthemr("light",layout = "minimal",type = "outer")
require(ggplot2)
ggplot(dfy,aes(x=`factor(time)`,y=mean,group=`factor(initial_dose)`))+
  geom_point(aes(colour=`factor(initial_dose)`,size=n))+
  geom_path(aes(colour=`factor(initial_dose)`))+
  #  geom_smooth(aes(colour=`factor(initial_dose)`),method="loess" )+
  facet_grid(`factor(behaviour)`~`factor(parent)`)+
  theme(axis.text.x=element_text(angle=45,hjust=1),panel.spacing.y = unit(1,"lines"))+
  xlab("")+ylab("mean level of behaviour")
## Scale for 'fill' is already present. Adding another scale for 'fill', which
## will replace the existing scale.
## Warning: Removed 26 rows containing missing values (geom_point).
## Warning: Removed 26 row(s) containing missing values (geom_path).
## geom_path: Each group consists of only one observation. Do you need to adjust
## the group aesthetic?
## geom_path: Each group consists of only one observation. Do you need to adjust
## the group aesthetic?

Parents correlation stratified by cohabiting

Correlation of maternal and paternal behaviour stratified by if live together or not

require(reshape2)
cors_lt <- cor(df1[df1$covs_partner_lives_with_mother_prenatal==1,vars[-((length(vars)-5):length(vars))]],use="pairwise.complete")
cors_nlt <- cor(df1[df1$covs_partner_lives_with_mother_prenatal==0,vars[-((length(vars)-5):length(vars))]],use="pairwise.complete")
cors_melt_lt <- melt(cors_lt)
cors_melt_lt$live_together <-"parents live together"
cors_melt_nlt <- melt(cors_nlt)
cors_melt_nlt$live_together <-"parents live apart"

cors_melt <- bind_rows(cors_melt_lt,cors_melt_nlt)

ggplot(cors_melt[grepl("smoking|alcohol|physact|caffeine",cors_melt$Var1)&grepl("smoking|alcohol|physact|caffeine",cors_melt$Var2),],aes(x=Var1,y=Var2,fill=value,label=round(value,2)))+
  geom_tile()+
  #  geom_text(colour="white")+
  theme(axis.text.x = element_text(angle = 45,hjust=1))+
  scale_fill_gradient(low="white",high="darkblue")+
  facet_grid(.~live_together)

Parents reduction at same time if cohabiting

Comparisons of maternal and paternal health behaviours, showing if reduced at the same time, stratified by if they lived together or apart

df7 <- cors_melt[grepl("smoking|alcohol|physact|caffeine",cors_melt$Var1)&grepl("smoking|alcohol|physact|caffeine",cors_melt$Var2),]

df7$time1 <- NA
df7$time1[grepl(df7$Var1,pattern="preconception")]<-"pre-conception"
df7$time1[grepl(df7$Var1,pattern="firsttrim|start")]<-"first trimester"
df7$time1[grepl(df7$Var1,pattern="secondtrim")]<-"second trimester"
df7$time1[grepl(df7$Var1,pattern="thirdtrim")]<-"third trimester"
df7$time1[grepl(df7$Var1,pattern="postnatal")]<-"postnatal"

df7$time2 <- NA
df7$time2[grepl(df7$Var2,pattern="preconception")]<-"pre-conception"
df7$time2[grepl(df7$Var2,pattern="firsttrim|start")]<-"first trimester"
df7$time2[grepl(df7$Var2,pattern="secondtrim")]<-"second trimester"
df7$time2[grepl(df7$Var2,pattern="thirdtrim")]<-"third trimester"
df7$time2[grepl(df7$Var2,pattern="postnatal")]<-"postnatal"
df7$sametime <- ifelse(df7$time1==df7$time2,"yes","no")

df7$behaviour1 <- unlist(lapply(strsplit(as.character(df7$Var1),split="_"),"[[",1))
df7$behaviour2 <- unlist(lapply(strsplit(as.character(df7$Var2),split="_"),"[[",1))
df7$samebehaviour <- ifelse(df7$behaviour1==df7$behaviour2,"yes","no")

df7$parent1 <- ifelse(grepl("mother",df7$Var1),"mother","partner")
df7$parent2 <- ifelse(grepl("mother",df7$Var2),"mother","partner")
df7$sameparent <- ifelse(df7$parent1==df7$parent2,"yes","no")

df8 <- df7[df7$sameparent=="no"&df7$samebehaviour=="yes"&df7$sametime=="yes",]
df8$time1 <- factor(df8$time1,ordered=T, levels=c("pre-conception","first trimester","second trimester","third trimester","postnatal"))

ggplot(df8,aes(x=1,y=value,fill=live_together))+
  geom_col(position = "dodge")+
  facet_grid(behaviour1~time1)

Parents reduction if other parent reduces behaviour also

Comparisons of maternal and paternal health behaviours, showing if reduced at the same time, stratified by if other parent reduced behaviour

  #New variables to include:
dat$smoking_father_change<-NA
dat$smoking_father_change[dat$smoking_father_startpregnancy_binary==0 & (dat$smoking_father_secondtrim_binary==1|dat$smoking_father_thirdtrim_binary==1)]<-1 #increased after start preg 
dat$smoking_father_change[dat$smoking_father_startpregnancy_binary==1 & (dat$smoking_father_secondtrim_binary==0|dat$smoking_father_thirdtrim_binary==0)]<-0 #decreased after start preg

#mums decrease smoking after start preg
dat$smoking_mother_change<-NA
dat$smoking_mother_change[dat$smoking_mother_firsttrim_binary==0 & (dat$smoking_mother_secondtrim_binary==1|dat$smoking_mother_thirdtrim_binary==1)]<-"Increased" #increased after start preg 
dat$smoking_mother_change[dat$smoking_mother_firsttrim_binary==1 & (dat$smoking_mother_secondtrim_binary==0|dat$smoking_mother_thirdtrim_binary==0)]<-"Decreased" #decreased after start preg




ggplot(dat, aes(x = smoking_mother_change,y = smoking_father_change,fill=smoking_father_change==1), position = "dodge")+
  ylim(0,125)+
labs(fill="Partner increased behaviour")+
labs(title = "ALSPAC")+
  xlab("Mum increased smoking during pregnancy")

## Works (back up if change above):
ggplot(data = dat) + 
  geom_bar(mapping = aes(x = smoking_mother_change, fill=smoking_father_change==1), position = "dodge")+
  ylim(0,125)+
labs(fill="Partner increased behaviour")+
labs(title = "ALSPAC")+
  xlab("Mum increased smoking during pregnancy")
## Warning: Removed 4 rows containing missing values (geom_bar).

###

?
  dat$time1<-df7$time1

df1<-mutate_all(dat[,c("aln",vars)], function(x) as.numeric(x))
df1_melt <- melt(df1[,grepl("aln|smoking|physact|caff|alco",names(df1))],id.vars = "aln")

BiB

Load data

This is taken from the BiB dataset created for EPoCH

rm(list=ls()) #clear previous cohorts data

dat <- readRDS("/Volumes/MRC-IEU-research/projects/ieu2/p5/015/working/data/bib/bib_pheno.rds")

Select variables

vars<-
  c(
    
   # "physact_mother_secondtrim_ordinal",
    #"physact_mother_thirdtrim_ordinal",
    #"physact_father_secondtrim_ordinal",
    #  "physact_mother_postnatal_ordinal", (not included as not directly comparable)
    
    # "caffeine_mother_total_firsttrim_ordinal",
   # "caffeine_mother_total_secondtrim_ordinal",
    "caffeine_mother_total_thirdtrim_ordinal",
  #  "caffeine_mother_total_postnatal_ordinal",
   # "caffeine_father_total_secondtrim_ordinal",
    
    "smoking_mother_preconception_ordinal",
    "smoking_mother_firsttrim_ordinal",
    "smoking_mother_secondtrim_ordinal",
    "smoking_mother_thirdtrim_ordinal",
   # "smoking_mother_postnatal_ordinal",
    
  #  "smoking_father_startpregnancy_ordinal",
   # "smoking_father_secondtrim_ordinal",
    "smoking_father_thirdtrim_ordinal",
  #  "smoking_father_postnatal_ordinal",
    
    "alcohol_mother_preconception_ordinal",
    "alcohol_mother_firsttrim_ordinal",
    "alcohol_mother_secondtrim_ordinal",
    "alcohol_mother_thirdtrim_ordinal",
   # "alcohol_mother_postnatal_ordinal",
    
  #  "alcohol_father_preconception_ordinal",
   # "alcohol_father_secondtrim_ordinal",
    "alcohol_father_thirdtrim_ordinal",
  #  "alcohol_father_postnatal_ordinal",
    
  #  "covs_occup_mother",
    "covs_occup_father",
    
    "covs_edu_mother",
    "covs_edu_father",
    
   # "covs_occup_mother_highestlowest_binary",
    "covs_occup_father_highestlowest_binary",
    "covs_edu_mother_highestlowest_binary",
    "covs_edu_father_highestlowest_binary",
    
    "covs_parity_mother_binary",
    "covs_partner_lives_with_mother_prenatal"
  )

Correlations

! To do format plots

Parents correlation

This shows how maternal and paternal health behaviours are correlated with each other and SEP

#install.packages("corrplot")

require(dplyr)
require(corrplot)
df1<-mutate_all(dat[,vars], function(x) as.numeric(x))
cors <- cor(df1,use="pairwise.complete")
corrplot(cors,tl.cex=0.5,method="color",tl.col="black",order="hclust",addrect=5)

Parents change over time

This shows how parental behaviour changes during and after pregnancy

#chaneged aln to PregnancyID for BiB (this should be right and not MotherID)
require(reshape2)
df1<-mutate_all(dat[,c("BiBMotherID",vars)], function(x) as.numeric(x))
## Warning in (function (x) : NAs introduced by coercion
df1_melt <- melt(df1[,grepl("BiBMotherID|smoking|caff|alco",names(df1))],id.vars = "BiBMotherID")

df1_melt$time <- NA
df1_melt$time[grepl(df1_melt$variable,pattern="preconception")]<-"pre-conception"
df1_melt$time[grepl(df1_melt$variable,pattern="firsttrim|start")]<-"first trimester"
df1_melt$time[grepl(df1_melt$variable,pattern="secondtrim")]<-"second trimester"
df1_melt$time[grepl(df1_melt$variable,pattern="thirdtrim")]<-"third trimester"
df1_melt$time[grepl(df1_melt$variable,pattern="postnatal")]<-"postnatal"
df1_melt$time <- factor(df1_melt$time,ordered=T, levels=c("pre-conception","first trimester","second trimester","third trimester","postnatal"))

df1_melt$behaviour <- NA
df1_melt$behaviour[grepl(df1_melt$variable,pattern="caff")]<-"caffeine"
df1_melt$behaviour[grepl(df1_melt$variable,pattern="alcohol")]<-"alcohol"
df1_melt$behaviour[grepl(df1_melt$variable,pattern="smoking")]<-"smoking"
#df1_melt$behaviour[grepl(df1_melt$variable,pattern="physac")]<-"physical activity"
df1_melt$behaviour <- factor(df1_melt$behaviour,ordered=T, levels=c("smoking","alcohol","caffeine"))

df1_melt$parent <- NA
df1_melt$parent[grepl(df1_melt$variable,pattern="mother")]<-"mother"
df1_melt$parent[grepl(df1_melt$variable,pattern="father")]<-"partner"

x <- droplevels(df1_melt[order(df1_melt$BiBMotherID,df1_melt$time,df1_melt$parent,df1_melt$behaviour),])
x<-x[is.na(x$value)==F,]
x <- x[duplicated(x[,c("parent","behaviour","BiBMotherID")])==F,]

names(x)[3]<-"initial_dose"
names(x)[2]<-"initial_var"

dfx <- merge(x[,c("BiBMotherID","parent","behaviour","initial_dose","initial_var")],df1_melt,by=c("BiBMotherID","parent","behaviour"),all=T)

dfy <- 
  dfx %>%
  group_by(factor(initial_dose),factor(behaviour),factor(time),factor(parent)) %>%
  dplyr::summarise(mean=mean(value,na.rm=T),sd=sd(value,na.rm=T),n=n())
## `summarise()` has grouped output by 'factor(initial_dose)', 'factor(behaviour)', 'factor(time)'. You can override using the `.groups` argument.
#devtools::install_github('Mikata-Project/ggthemr')

ggthemr::ggthemr("light",layout = "minimal",type = "outer")
require(ggplot2)
ggplot(dfy,aes(x=`factor(time)`,y=mean,group=`factor(initial_dose)`))+
  geom_point(aes(colour=`factor(initial_dose)`,size=n))+
  geom_path(aes(colour=`factor(initial_dose)`))+
  #  geom_smooth(aes(colour=`factor(initial_dose)`),method="loess" )+
  facet_grid(`factor(behaviour)`~`factor(parent)`)+
  theme(axis.text.x=element_text(angle=45,hjust=1),panel.spacing.y = unit(1,"lines"))+
  xlab("")+ylab("mean level of behaviour")
## Scale for 'fill' is already present. Adding another scale for 'fill', which
## will replace the existing scale.
## geom_path: Each group consists of only one observation. Do you need to adjust
## the group aesthetic?
## geom_path: Each group consists of only one observation. Do you need to adjust
## the group aesthetic?
## geom_path: Each group consists of only one observation. Do you need to adjust
## the group aesthetic?

Parents correlation stratified by cohabiting

Correlation of maternal and paternal behaviour stratified by if live together or not

require(reshape2)
cors_lt <- cor(df1[df1$covs_partner_lives_with_mother_prenatal==1,vars[-((length(vars)-5):length(vars))]],use="pairwise.complete")
cors_nlt <- cor(df1[df1$covs_partner_lives_with_mother_prenatal==0,vars[-((length(vars)-5):length(vars))]],use="pairwise.complete")
cors_melt_lt <- melt(cors_lt)
cors_melt_lt$live_together <-"parents live together"
cors_melt_nlt <- melt(cors_nlt)
cors_melt_nlt$live_together <-"parents live apart"

cors_melt <- bind_rows(cors_melt_lt,cors_melt_nlt)

ggplot(cors_melt[grepl("smoking|alcohol|physact|caffeine",cors_melt$Var1)&grepl("smoking|alcohol|physact|caffeine",cors_melt$Var2),],aes(x=Var1,y=Var2,fill=value,label=round(value,2)))+
  geom_tile()+
  #  geom_text(colour="white")+
  theme(axis.text.x = element_text(angle = 45,hjust=1))+
  scale_fill_gradient(low="white",high="darkblue")+
  facet_grid(.~live_together)

Parents reduction at same time if cohabiting

Comparisons of maternal and paternal health behaviours, showing if reduced at the same time, stratified by if they lived together or apart

df7 <- cors_melt[grepl("smoking|alcohol|caffeine",cors_melt$Var1)&grepl("smoking|alcohol|caffeine",cors_melt$Var2),]

df7$time1 <- NA
df7$time1[grepl(df7$Var1,pattern="preconception")]<-"pre-conception"
df7$time1[grepl(df7$Var1,pattern="firsttrim|start")]<-"first trimester"
df7$time1[grepl(df7$Var1,pattern="secondtrim")]<-"second trimester"
df7$time1[grepl(df7$Var1,pattern="thirdtrim")]<-"third trimester"
df7$time1[grepl(df7$Var1,pattern="postnatal")]<-"postnatal"

df7$time2 <- NA
df7$time2[grepl(df7$Var2,pattern="preconception")]<-"pre-conception"
df7$time2[grepl(df7$Var2,pattern="firsttrim|start")]<-"first trimester"
df7$time2[grepl(df7$Var2,pattern="secondtrim")]<-"second trimester"
df7$time2[grepl(df7$Var2,pattern="thirdtrim")]<-"third trimester"
df7$time2[grepl(df7$Var2,pattern="postnatal")]<-"postnatal"
df7$sametime <- ifelse(df7$time1==df7$time2,"yes","no")

df7$behaviour1 <- unlist(lapply(strsplit(as.character(df7$Var1),split="_"),"[[",1))
df7$behaviour2 <- unlist(lapply(strsplit(as.character(df7$Var2),split="_"),"[[",1))
df7$samebehaviour <- ifelse(df7$behaviour1==df7$behaviour2,"yes","no")

df7$parent1 <- ifelse(grepl("mother",df7$Var1),"mother","partner")
df7$parent2 <- ifelse(grepl("mother",df7$Var2),"mother","partner")
df7$sameparent <- ifelse(df7$parent1==df7$parent2,"yes","no")

df8 <- df7[df7$sameparent=="no"&df7$samebehaviour=="yes"&df7$sametime=="yes",]
df8$time1 <- factor(df8$time1,ordered=T, levels=c("pre-conception","first trimester","second trimester","third trimester","postnatal"))

ggplot(df8,aes(x=1,y=value,fill=live_together))+
  geom_col(position = "dodge")+
  facet_grid(behaviour1~time1)

MCS

rm(list=ls())

Load data

This is taken from the MCS dataset created for EPoCH

dat<-readRDS("/Volumes/MRC-IEU-research/projects/ieu2/p5/015/working/data/mcs/mcs_pheno.rds")

Select variables

vars<-
  c(
    
   # "physact_mother_secondtrim_ordinal",
    #"physact_mother_thirdtrim_ordinal",
    #"physact_father_secondtrim_ordinal",
    #  "physact_mother_postnatal_ordinal", (not included as not directly comparable)
    
   # "caffeine_mother_total_firsttrim_ordinal",
  #  "caffeine_mother_total_secondtrim_ordinal",
   # "caffeine_mother_total_thirdtrim_ordinal",
  #  "caffeine_mother_total_postnatal_ordinal",
  #  "caffeine_father_total_secondtrim_ordinal",
    
    "smoking_mother_preconception_ordinal",
    "smoking_mother_firsttrim_ordinal",
    "smoking_mother_secondtrim_ordinal",
    "smoking_mother_thirdtrim_ordinal",
    "smoking_mother_postnatal_ordinal",
    
   # "smoking_father_startpregnancy_ordinal",
        "smoking_father_firsttrim_ordinal",
    "smoking_father_secondtrim_ordinal",
    "smoking_father_thirdtrim_ordinal",
    "smoking_father_postnatal_ordinal",
    
    #"alcohol_mother_preconception_ordinal",
    #"alcohol_mother_firsttrim_ordinal",
   # "alcohol_mother_secondtrim_ordinal",
  #  "alcohol_mother_thirdtrim_ordinal",
    "alcohol_mother_postnatal_ordinal",
   #  We also have an ever pregnancy just for mums "alcohol_mother_ever_pregnancy_ordinal",

   # "alcohol_father_preconception_ordinal",
  #  "alcohol_father_secondtrim_ordinal",
   # "alcohol_father_thirdtrim_ordinal",
    #"alcohol_father_postnatal_ordinal",
  #  "alcohol_father_postnatal_ordinal"
  
    "covs_occup_mother",
    "covs_occup_father",
    
    "covs_edu_mother",
    "covs_edu_father",
    
    "covs_occup_mother_highestlowest_binary",
    "covs_occup_father_highestlowest_binary",
    "covs_edu_mother_highestlowest_binary",
    "covs_edu_father_highestlowest_binary",
    
   # "covs_parity_mother_binary",
    "covs_partner_lives_with_mother_prenatal"
  )

Correlations

! To do format plots

Parents correlation

This shows how maternal and paternal health behaviours are correlated with each other and SEP

#install.packages("corrplot")

require(dplyr)
require(corrplot)

df1<-mutate_all(dat[,vars], function(x) as.numeric(x))
cors <- cor(df1,use="pairwise.complete")
corrplot(cors,tl.cex=0.5,method="color",tl.col="black",order="hclust",addrect=5)

Parents change over time

This shows how parental behaviour changes during and after pregnancy

require(reshape2)
df1<-mutate_all(dat[,c("mcsid",vars)], function(x) as.numeric(x))
## Warning in (function (x) : NAs introduced by coercion
df1_melt <- melt(df1[,grepl("mcsid|smoking|physact|caff|alco",names(df1))],id.vars = "mcsid")

df1_melt$time <- NA
df1_melt$time[grepl(df1_melt$variable,pattern="preconception")]<-"pre-conception"
df1_melt$time[grepl(df1_melt$variable,pattern="firsttrim|start")]<-"first trimester"
df1_melt$time[grepl(df1_melt$variable,pattern="secondtrim")]<-"second trimester"
df1_melt$time[grepl(df1_melt$variable,pattern="thirdtrim")]<-"third trimester"
df1_melt$time[grepl(df1_melt$variable,pattern="postnatal")]<-"postnatal"
df1_melt$time <- factor(df1_melt$time,ordered=T, levels=c("pre-conception","first trimester","second trimester","third trimester","postnatal"))

df1_melt$behaviour <- NA
#df1_melt$behaviour[grepl(df1_melt$variable,pattern="caff")]<-"caffeine"
df1_melt$behaviour[grepl(df1_melt$variable,pattern="alcohol")]<-"alcohol"
df1_melt$behaviour[grepl(df1_melt$variable,pattern="smoking")]<-"smoking"
#df1_melt$behaviour[grepl(df1_melt$variable,pattern="physac")]<-"physical activity"
df1_melt$behaviour <- factor(df1_melt$behaviour,ordered=T, levels=c("smoking","alcohol","caffeine","physical activity"))

df1_melt$parent <- NA
df1_melt$parent[grepl(df1_melt$variable,pattern="mother")]<-"mother"
df1_melt$parent[grepl(df1_melt$variable,pattern="father")]<-"partner"

x <- droplevels(df1_melt[order(df1_melt$mcsid,df1_melt$time,df1_melt$parent,df1_melt$behaviour),])
x<-x[is.na(x$value)==F,]
x <- x[duplicated(x[,c("parent","behaviour","mcsid")])==F,]

names(x)[3]<-"initial_dose"
names(x)[2]<-"initial_var"

dfx <- merge(x[,c("mcsid","parent","behaviour","initial_dose","initial_var")],df1_melt,by=c("mcsid","parent","behaviour"),all=T)

dfy <- 
  dfx %>%
  group_by(factor(initial_dose),factor(behaviour),factor(time),factor(parent)) %>%
  dplyr::summarise(mean=mean(value,na.rm=T),sd=sd(value,na.rm=T),n=n())
## `summarise()` has grouped output by 'factor(initial_dose)', 'factor(behaviour)', 'factor(time)'. You can override using the `.groups` argument.
ggthemr::ggthemr("light",layout = "minimal",type = "outer")
require(ggplot2)
ggplot(dfy,aes(x=`factor(time)`,y=mean,group=`factor(initial_dose)`))+
  geom_point(aes(colour=`factor(initial_dose)`,size=n))+
  geom_path(aes(colour=`factor(initial_dose)`))+
  #  geom_smooth(aes(colour=`factor(initial_dose)`),method="loess" )+
  facet_grid(`factor(behaviour)`~`factor(parent)`)+
  theme(axis.text.x=element_text(angle=45,hjust=1),panel.spacing.y = unit(1,"lines"))+
  xlab("")+ylab("mean level of behaviour")
## Scale for 'fill' is already present. Adding another scale for 'fill', which
## will replace the existing scale.
## geom_path: Each group consists of only one observation. Do you need to adjust
## the group aesthetic?

Parents correlation stratified by cohabiting

Correlation of maternal and paternal behaviour stratified by if live together or not

require(reshape2)
cors_lt <- cor(df1[df1$covs_partner_lives_with_mother_prenatal==1,vars[-((length(vars)-5):length(vars))]],use="pairwise.complete")
cors_nlt <- cor(df1[df1$covs_partner_lives_with_mother_prenatal==0,vars[-((length(vars)-5):length(vars))]],use="pairwise.complete")
cors_melt_lt <- melt(cors_lt)
cors_melt_lt$live_together <-"parents live together"
cors_melt_nlt <- melt(cors_nlt)
cors_melt_nlt$live_together <-"parents live apart"

cors_melt <- bind_rows(cors_melt_lt,cors_melt_nlt)

ggplot(cors_melt[grepl("smoking|alcohol|physact|caffeine",cors_melt$Var1)&grepl("smoking|alcohol|physact|caffeine",cors_melt$Var2),],aes(x=Var1,y=Var2,fill=value,label=round(value,2)))+
  geom_tile()+
  #  geom_text(colour="white")+
  theme(axis.text.x = element_text(angle = 45,hjust=1))+
  scale_fill_gradient(low="white",high="darkblue")+
  facet_grid(.~live_together)

Parents reduction at same time if cohabiting

Comparisons of maternal and paternal health behaviours, showing if reduced at the same time, stratified by if they lived together or apart

df7 <- cors_melt[grepl("smoking|alcohol|physact|caffeine",cors_melt$Var1)&grepl("smoking|alcohol|physact|caffeine",cors_melt$Var2),]

df7$time1 <- NA
df7$time1[grepl(df7$Var1,pattern="preconception")]<-"pre-conception"
df7$time1[grepl(df7$Var1,pattern="firsttrim|start")]<-"first trimester"
df7$time1[grepl(df7$Var1,pattern="secondtrim")]<-"second trimester"
df7$time1[grepl(df7$Var1,pattern="thirdtrim")]<-"third trimester"
df7$time1[grepl(df7$Var1,pattern="postnatal")]<-"postnatal"

df7$time2 <- NA
df7$time2[grepl(df7$Var2,pattern="preconception")]<-"pre-conception"
df7$time2[grepl(df7$Var2,pattern="firsttrim|start")]<-"first trimester"
df7$time2[grepl(df7$Var2,pattern="secondtrim")]<-"second trimester"
df7$time2[grepl(df7$Var2,pattern="thirdtrim")]<-"third trimester"
df7$time2[grepl(df7$Var2,pattern="postnatal")]<-"postnatal"
df7$sametime <- ifelse(df7$time1==df7$time2,"yes","no")

df7$behaviour1 <- unlist(lapply(strsplit(as.character(df7$Var1),split="_"),"[[",1))
df7$behaviour2 <- unlist(lapply(strsplit(as.character(df7$Var2),split="_"),"[[",1))
df7$samebehaviour <- ifelse(df7$behaviour1==df7$behaviour2,"yes","no")

df7$parent1 <- ifelse(grepl("mother",df7$Var1),"mother","partner")
df7$parent2 <- ifelse(grepl("mother",df7$Var2),"mother","partner")
df7$sameparent <- ifelse(df7$parent1==df7$parent2,"yes","no")

df8 <- df7[df7$sameparent=="no"&df7$samebehaviour=="yes"&df7$sametime=="yes",]
df8$time1 <- factor(df8$time1,ordered=T, levels=c("pre-conception","first trimester","second trimester","third trimester","postnatal"))

ggplot(df8,aes(x=1,y=value,fill=live_together))+
  geom_col(position = "dodge")+
  facet_grid(behaviour1~time1)